Upload
nguyenkhanh
View
222
Download
0
Embed Size (px)
Citation preview
Loughborough UniversityInstitutional Repository
Surface cleaningtechnologies for the removalof crosslinked epoxide resin
This item was submitted to Loughborough University's Institutional Repositoryby the/an author.
Citation: LITCHFIELD, R.E., CRITCHLOW, G.W. and WILSON, S., 2006.Surface cleaning technologies for the removal of crosslinked epoxide resin. In-ternational Journal of Adhesion and Adhesives, 26(5), pp. 295-303
Additional Information:
• This article has been accepted for publication in the journal, InternationalJournal of Adhesion and Adhesives [ c© Elsevier]. The definitive version isavailable at: http://www.sciencedirect.com/science/journal/01437496.
Metadata Record: https://dspace.lboro.ac.uk/2134/861
Publisher: c© Elsevier
Please cite the published version.
Surface cleaning technologies for the removal of crosslinked epoxide resin
R.E.Litchfield1, G.W.Critchlow1* and S.Wilson2
1Institute of Surface Science and Technology, IPTME, Loughborough University,
Loughborough, Leicestershire, LE11 3TU, UK 2Bombardier Aerospace/Short Bros., Airport Road, Belfast, Northern Ireland, BT3 9DZ, UK
*author to whom correspondence should be addressed, [email protected]
Keywords: epoxides, metals, abhesion/non-stick.
Abstract
This study provides details of the use of laser ablation and sodium hydride cleaning processes for
the removal of crosslinked epoxide and other residues from resin transfer moulding (RTM) tool
substrates, as used in the aerospace industry. The requirement for removal of such contamination is
so that the mould can be re-used, following the subsequent application of an external release agent.
These tools are, typically, fabricated from steel, nickel or CFRP composite materials; this paper
focuses on the use of nickel substrates. The requirement to clean large surface areas quickly to
satisfy commercial restraints, compromises the degree of absolute cleanliness that can be obtained.
However, in applications where cleaning time is not a constraint, laser cleaning can be a very gentle
and efficient process; typically Nd:YAG lasers find application in this area. In contrast, high power
lasers are desirable for industrial scale applications where large areas need to be cleaned quickly. In
this instance pulsed CO2 lasers can be used. The use of sodium hydride was also found to be highly
successful in removing crosslinked organic contamination providing that suitable hard rinse and
drying operations were also carried out.
1. Introduction
The removal of organic material such as lubricating oils or other processing aids, for example
release agents, is routinely carried out as part of most multi-stage metal finishing processes1.
Commonly, aqueous-based or vapour phase solvent media are used which can effectively remove
relatively weakly bound organic materials. The aqueous-based media are often combined with
agitation including ultrasonics for maximum speed and efficiency.
The current study is, however, concerned with the removal of fully crosslinked organic material
from metal surfaces. It is not possible to effectively remove such material using the conventional
cleaning procedures referred to above. In particular, this paper focuses on the removal of cured
epoxide resin from nickel surfaces using laser ablation. Some consideration of the choice of laser
type and laser-coating interactions is contained in this paper as these factors help to explain the
experimental results. An alternative process, namely, sodium hydride cleaning will also be
discussed although in limited detail. The application of this technology is in the aerospace industry
where metal mould tools are used in the production of resin transfer moulding (RTM) components.
Occasionally, the tooling requires cleaning to remove residual epoxides from the casting process so
that the tools can be re-used following the application of an external mould release system. An
additional requirement for a successful cleaning procedure is therefore to create a surface that can
accommodate the application of an organic coating. Cured epoxide resins are both cohesively
strong and may be attached via mechanical interlocking or chemisorbed onto the nickel.
The removal of residual epoxides from mould tools can be carried out using mechanical methods,
for example, sanding or grit-blasting. These methods are both extremely time consuming and
induce significant amounts of surface physical damage rendering the moulds more liable to
contamination on subsequent usage. Alternatively, industrial scale cleaning solutions may be used
incorporating methylene chloride. Current environmental and health and safety legislation are
acting as driving forces to identify new cleaning methods that obviate the need for this process.
Within the last decade laser cleaning has emerged as a viable solution to such problems. The use of
lasers in conservation and art restoration2 is a mature technology but the use of lasers for industrial
scale applications, where the time taken to clean and capital equipment costs are critical factors, is
only now being fully addressed. The use of laser processing to promote adhesion by surface
desorption, deposition and texturing is also recognised but such processes are not directly relevant
to the present study3-6.
Excimer lasers operating at ultra-violet (uv) wavelengths are successful at processing many organic
polymers and are effective in laser cleaning but the relatively high absorptivity of metals at uv
wavelengths is a problem where heating or melting effects upon the metal substrate have to be
considered. In the current application little or no interaction with the substrate is desirable. A recent
novel and successful application of laser cleaning utilising a Nd: YAG (Neodymium: Yttrium
Aluminium Garnet) laser has been to clean leaves from railway tracks7. Although lasers can be
used to harden metals by introducing compressive surface stresses, this does not occur under the
typical operating conditions used for laser cleaning. In this example, metallurgical tests have shown
that no increase in the martensite phase of the steel or other detrimental thermal damage to the
railway tracks occurred after more than five billion pulses.
Other applications of laser cleaning include tyre mould refurbishment8, although little has been
published of relevance to larger industrial scale cleaning of metal moulds and, in particular, to the
removal of cured resin from such tooling. Robotically controlled solid-state lasers used in tyre
mould cleaning can clean approximately one square metre of surface area within 45 to 60 minutes,
representing a very significant improvement on conventional abrasive cleaning methods.
In addition, powerful multi-kilowatt (kW) carbon dioxide (CO2) laser systems have been built by
the United States military and have been applied to the paint stripping of aircraft. These have
reportedly achieved paint stripping rates of 30 to 60 m2 per hour. Commercial systems have been
developed operating up to 2 kW and have proved highly successful9. Stripping of paint from
aircraft is periodically necessary to allow metallurgical inspection for metal fatigue. Both these
examples represent expensive cleaning solutions and require bespoke designs. To be competitive
with other cleaning methods such as dry ice pellet cleaning10,11 or high-pressure water jet systems,
any laser cleaning solution must be very reliable and competitively priced.
The efficiency of mould release is another factor of importance in the aerospace industry and the
adhesion of any applied mould release agent is reliant on obtaining a clean tool surface. Many
studies have shown that surface cleanliness promotes good coating adhesion.
Laser cleaning of large surface areas presently utilises TEA (Transversely Excited Atmosphere)
CO2 lasers possessing multi-kilowatts energy outputs. These are pulsed lasers that can clean large
areas quickly. Carbon dioxide lasers are an obvious choice since the laser output wavelength of
10.6 micrometres (µm) couples very effectively to the molecular vibrational modes of many organic
materials causing them to heat up and degrade. An organic resin should, therefore, readily absorb
energy at this wavelength. Metal substrates are also highly reflective to this laser wavelength with
the consequence that the cleaning process is self-limiting with only the organic contaminant being
removed. TEA CO2 lasers require separate gas supplies to operate. Sealed radio frequency excited
CO2 lasers are an alternative that do not require gas supplies and their reduced physical size makes
them more suitable for robotic mounting. These systems though are not as powerful as TEA CO2
lasers and the intended market for them is biased towards materials processing rather than cleaning
applications. Shorter wavelength lasers such the Nd:YAG solid state laser are also available for
cleaning and find greatest application in art restoration work but the reflectivity of metals falls off at
shorter wavelengths so that there is a greater chance that some substrate damage may occur when
they are used to remove coatings from metals.
Most applications of CO2 lasers utilise CNC tables to move the work-piece requiring cleaning with
the laser being either static or mounted on a robot arm; with appropriate control over the arm the
laser is scanned over the area to be cleaned. Where the cleaning of very large moulding tools is
considered, the workpiece cannot be moved and some form of automation is required to scan the
laser over the area to be cleaned. The physical size of a TEA CO2 laser would suggest that this
option would be costly to implement and not very flexible. Aside from the cost issues involved in
realising an industrial scale laser cleaning solution with CO2 lasers, the technology has some other
disadvantages. At present, there exist few materials that will transmit the 10.6 µm wavelength
without significant absorption. The main exceptions are zinc selenide or germanium and focussing
optics have to be produced from these rather exotic materials. Highly polished mirrors made from
gold or copper can also be used because of the high reflectivity of metals at 10.6 µm. Unlike CO2
lasers used for cutting and welding processes, where optics focus the laser light down to very small
spot sizes, laser cleaning requires defocused beams covering an area of a few square centimetres.
Ideally, the energy distribution across such beams should be uniform but this is difficult to achieve
because of the way high power CO2 lasers are designed. Measuring laser energy densities (or
fluences) for intense lasers poses experimental problems since conventional power meters used for
most other laser power measurements are very easily damaged by high-energy pulses and very
expensive to replace. Such measurements are desirable to optimise any cleaning process but are
difficult to make and estimated measurement errors of ±10% or more are common. Other critical
factors are the pulse duration and repetition rate. A minimum (or threshold) intensity is required to
thermally ablate any contaminant layer and this is a function of the material composition.
Despite a greater propensity to cause thermal damage on metal substrates due to the reduced
reflectivity, the use of Nd:YAG lasers, operating at an output wavelength of 1.06 µm, offers some
distinct advantages. Glass fibre optic beam delivery from the laser source to the work piece is
possible because there is very little absorption by the glass at this wavelength. This greatly
enhances the flexibility of the cleaning process since a skilled operator can now manoeuvre a hand
held laser cleaning head over the area to be cleaned and if necessary simply increase the fluence to
remove stubborn contamination. The use of a solid-state laser also makes the laser inherently more
reliable and free from operating faults. The technology of these lasers is well developed. Unlike
TEA CO2 lasers, which are physically bulky systems requiring gas and three phase electrical power
supplies the size of Nd:YAG systems has been steadily reduced and current small footprint units are
marketed. These advantages are made at the sacrifice of cleaning rates which are slower than using
CO2 lasers but still competitive with other non-laser cleaning methods. In most other respects
Nd:YAG lasers offer the most cost effective laser cleaning solution.
Paint stripping applications are also possible using relatively small, compact diode laser arrays
where power outputs can be as high as 5 kW. Using oxygen assist gas, it has been reported that a
120 W diode laser was capable of removing thick coatings of white, chlorinated rubber paint from
concrete surfaces at cleaning rates of several square metres per hour. However, an oxygen assist
gas was used to affect a controlled combustion and it is conjectured that this would be unsuitable
for coating removal from most metal substrates.
One drawback of laser cleaning is that it is largely a line-of-sight technique and has difficulty in
cleaning very complex shaped mould tooling. In such cases there is little alternative but to resort to
a chemical cleaning method where the tool is lowered into a bath of cleaning agent. In these
applications it is also desirable that the cleaning chemistry used removes silicones arising from
degraded mould release agents as well as crosslinked epoxide resin. The chemical inertness of
silicones requires aggressive and very searching cleaning agents that will not damage the metal
substrate. One technology that will fulfil these requirements is the use of a fused alkali bath. Aside
from the toxicity and high operating temperature of this process, no other solvents are involved and
so waste is kept to a minimum. There is, however, a practical size limitation on the tooling that is
amenable to this cleaning method. For reasons that will be discussed the process is best suited to
nickel or metals that are not susceptible to rapid rusting once a clean surface has been obtained.
DuPont developed a sodium hydride treatment in 194212 as a method for removing oxide scale and
casting sand residues from metal castings using a fused alkali bath, which would not cause any
damage to the casting. The process has been described in detail by Lightfoot13. Lightfoot applied
the sodium hydride process to the destruction of toxic chlorinated organic compounds.
The process itself requires the part to be immersed in a molten bath of anhydrous sodium hydroxide
maintained at 360°C. Hydrogen is passed through the melt to create a relatively dilute 2% solution
of sodium hydride. The sodium hydride is a very powerful reducing agent capable of removing
most metal oxides and exposing the bare metal. Refractory oxides such as those of aluminium,
titanium or silicon are not reduced but are removed by conversion to the corresponding elemental
silicates by the highly basic anhydrous melt. Unlike conventional acid pickling solutions the
substrate metal is not attacked by this highly basic medium provided the metal is immediately
quenched in a large excess of clean water upon removal from the melt. Many steels cleaned in this
way would rapidly rust in the time taken to move them to any oven for drying and so it is suggested
that the process is better suited to metals such as nickel. The very low surface tension of the melt
renders the treatment extremely searching penetrating into any hairline cracks or the interfacial
regions between any coating and the metal. Combined with the high temperature of the melt, and
net action on any organic coatings such as resins would be to carbonise them and pressure washing
is necessary to remove carbonised organic residues from the metal surface.
Sodium hydride cleaning has the advantage of being one of very few methods that will remove any
silicone contamination from a metal surface. Cleaning time is relatively quick but the process is
obviously hazardous although process controls and containment can be achieved.
In summary, the there are a number of technological issues which make the removal of fully
crosslinked epoxide resin from rough metal surfaces difficult to achieve. This is particularly
difficult if this is to be achieved with minimum interaction with the underlying substrate. In the
present study, two options: laser ablation, and; sodium hydride treatment will be considered for the
removal of epoxide resin from nickel substrates.
2. Experimental
2.1 Materials
Coupons of nickel plate, measuring 5 cm x 5 cm x 5 mm thickness, were used as substrates for the
bonding of epoxide resin. Surfaces were mechanically scoured using abrasive pads to produce a
roughness measurement (Ra) of 0.405µm. The following cleaning regime was then applied.
Coupons were initially doubly-degreased using ultrasonic agitation in super-purity acetone. These
were then immersed in boiling aqueous alkaline cleaner (Isoprep 44) with mechanical agitation for a
period of 30 minutes. Each sample was then rinsed in running hot tap water and transferred to a
fresh solution of the alkaline cleaner which was at room temperature. Samples were then subjected
to ultrasonic agitation for a further 30 minutes before each was again rinsed in hot tap water and
thoroughly dried in an oven at 120°C for 60 minutes.
An average water contact angle of 32° was measured for the nickel samples following this
treatment. This suggested the nickel samples were reasonably clean. Note that clean metal surfaces
attract a thin layer of contamination very quickly and measured contact angles rapidly increased to
50° and above. Surface roughness is also a factor affecting contact angles and the same cleaning
regime applied to a smoother stainless steel foil resulted in values between 10° and 20°.
The epoxide resin system used to coat the samples was Hexcel RTM6. This was warmed to reduce
its viscosity, a quantity was then thoroughly mixed with an equal volume of acetone which was
used as a solvent. This solution was then brushed onto cleaned metal coupons. After a single
application the coupon was cured for 4 hours in an oven at 180°C. thickness measurements from
coupons after cooling of the cured resin showed the resin thickness to be within the range 100 to
150 µm, the greater thickness occurring towards the edge of the samples.
2.2 Laser Cleaning
A study of the effects of CO2 lasers on resin coated nickel tooling was performed using a
Laserbrand L450 TEA CO2 laser, emitting 10.6 µm radiation in a 100 ns pulse of 2J output energy.
The unfocussed beam is approximately square with an area of 10cm2 which was focussed down to
1cm2 using a plano-convex ZnSe lens of focal length 30cm. At this area the fluence is great enough
to readily vaporise organic coatings. Coated metal samples were then positioned at various
distances from the focus to receive differing fluences. The samples were all mounted such that the
laser beam was incident normal to the surface of the coated samples. This was done to eliminate
angle of incidence as a variable and simplify evaluation. The number of pulses required to produce
a visible effect on the coating for a given fluence was noted. With smaller, much lighter samples a
sensitive mass balance could be used to monitor weight loss as the coating was removed but this
was considered to be impractical in this study.
Generally, it has been found that cleaning fluences range from 1 to 4 J.cm-2 depending on the
absorption characteristics of the coating. Higher fluences may be required for weakly absorbing
materials. Air breakdown occurs at fluences ~10 J.cm-2 and this situation is avoided to prevent
plasma formation, which produces many undesirable secondary effects. Sometimes these plasmas
are triggered by the non-removal or ineffective removal of ablated material. Since very high
temperatures are a characteristic of plasmas these can radiate sufficient energy to the substrate and
induce thermal damage.
The laser cleaning effects obtainable from the lower 1.060 µm wavelength radiation from Nd:YAG
lasers were investigated using a Spectron SL456 pulsed Nd:YAG. The laser utilises a temperature
stabilised Pockels cell to effect Q-switching, enabling the laser to produce a short, intense pulse of
energy (duration 13ns). Output power level and repetition rate were adjusted to give a maximum
output of 850mJ up to 1 0 Hz. Repetition rates up to 5 Hz were used giving pulse energies of
approximately 600mJ, this value being interpolated from the laser’s output calibration
measurement. The focussed laser beam was powerful enough to bore holes into nickel plate so
samples requiring cleaning were again positioned at varying distances from the laser focal point and
the coating removal effects noted until an optimum distance was found. This proved to be
approximately 50mm from laser focus. Using this degree of defocusing the beam diameter was
about 8mm and a single pulse was sufficient to detach cured resin from the metal substrate to which
it was bonded. Ideally, the laser beam used for a cleaning application should be spatially
homogenous with a “top-hat” shaped intensity distribution. Laboratory lasers generally do not meet
this requirement and in the case of a Nd:YAG laser, may have Gaussian beam profiles more suited
to non cleaning applications and materials processing. This consideration is particularly severe with
TEA CO2 lasers where the beam is never homogenous and there exists a distribution of fluence over
its area. This problem makes it difficult to correlate microstructure observed on treated samples
with an accurately known fluence. In the case of Nd:YAG lasers, the manufacturers of a
commercial Nd:YAG laser designed for cleaning applications (and possessing a homogenised
beam) were approached and some resin coated samples submitted for laser cleaning. These samples
were cleaned at a rate of 10cm2 per second using a fluence of 1J.cm2 with normal incidence of the
beam. The microstructural topography was found to be comparable to those obtained with the
Spectron laser.
2.3 Sodium Hydride Cleaning
Resin coated nickel tooling was immersed in a bath containing anhydrous sodium hydroxide with a
2% addition of sodium hydride maintained at 360°C for a period of 60 minutes. After removal the
tooling was thoroughly rinsed and stored in absolute ethanol to minimise subsequent oxidation prior
to analysis.
Results
3.1 Samples cleaned using TEA CO2 laser ablation
Figures 1a and 1b show macro photographs of a resin coated nickel plate in which strips of resin
have been removed using a laser fluence of 6 J.cm-2. When examined in the scanning electron
microscope (SEM), the ‘cleaned’ areas show thin residual patches of resin which are visible as dark
areas, see Figure 1c. Areas free from any such residue show a clean surface with no thermal
damage to the nickel substrate, see Figure 1d.
3.2 Samples cleaned using Nd:YAG laser ablation
Figures 2a and 2b show macro photographs of a resin coated nickel plate from where detachment of
the resin has occurred in an area treated with the Nd:YAG laser. A laser fluence of 1 J.cm-2 was
used. Figures 2c and 2d show SEM images of the cleaned substrate beneath the detached resin.
The substrate is exceptionally clean. The scratches originate from the abrasive scouring of the
samples prior to resin bonding. Figure 2d shows that some slight thermal damage has occurred due
to the greater absorption of the metal substrate for Nd:YAG radiation of 1.060 µm wavelength. It is
argued that this damage is inconsequential compared to that produced by abrasive cleaning
procedures.
Auger electron spectroscopy (AES) was used to determine the level of residual surface
contamination on the laser cleaned sample. Argon ion bombardment was used to determine sub-
surface compositions and an empirical etch rate was applied to enable depth scale calibration. It
was concluded from AES that the thickness of surface contamination following laser ablation of a
150 µm coating of resin was only approximately 5nm. This demonstrates the effectiveness of the
Nd:YAG process in this application. A water contact angle of 38° was obtained for the freshly
cleaned nickel surface following Nd:YAG laser removal of resin.
Where the contamination is resinous but does not form a continuously cohesive film, comprising
instead particulate matter with granular texture, laser cleaning using Nd:YAG is still effective.
Figure 2e shows that such contamination on nickel is removed leaving a bright metal surface. Upon
examination using scanning electron microscopy, the contrast between a cleaned and uncleaned
area is marked. The mechanism of detachment in this case is more complex and less easy to
interpret.
3.3 Samples cleaned using the sodium hydride process
Figure 3a is a macro photograph showing a resin coated nickel sample adjacent to one that has been
treated using the sodium hydride process. This process was shown to chemically reduce any
organic material present to a black char which is easily removed using a water jet provided this is
done immediately following treatment. Following such treatment a bright metal surface is revealed
and visual inspection shows no trace of any resin residue. Water contact angle measurements gave
an average angle of 40° but the surface rapidly contaminated by absorption from the atmosphere.
Examination of such a cleaned surface using scanning electron microscopy reveals only the
scratches from abrasive scouring prior to bonding of the resin, see Figures 3b and 3c.
By using energy dispersive X-ray microanalysis (EDXA), the elemental composition of the cleaned
surface is revealed, see Figure 3d, this shows only trace levels of elemental impurities below the
main intense peaks associated with nickel. The most likely source of these impurities is from
samples previously cleaned using the process since a commercial metal finishing bath was used.
Again, AES was used to confirm the surface cleanliness, in this case the thickness of the residual
surface contamination was found to be approximately 30nm. Although thicker than that obtained
by laser cleaning the level of contamination is slight and relatively inconsequential for industrial
applications.
4. Discussion
The mechanism of RTM6 epoxide coating removal varies depending on the pulse duration.
Considering a general coating such as a multi-layer paint, irradiation by high-energy pulses in the
order of a few milliseconds will burn the coating but may also affect the substrate if there is
sufficient for heat conduction to occur by thermal diffusion.
Lasers used for cutting and welding of metals operate with very long pulse durations of several
hundred milliseconds and at high repetition rates. In comparison, shorter nanosecond pulses will
not allow any heat transfer to occur to the substrate. When incident on a coating above threshold
fluence, these short intense pulses will instantly vaporise the surface layer creating a rapidly
expanding cloud resulting in a mechanical shock wave that propagates through the coating and is
reflected back from the substrate. The surface pressures created are in the order of kilobars and are
more than sufficient to spontaneously eject whole layers of surface coating, apparently overcoming
any interfacial forces associated with the adhesion of the coating. The cleaning mechanism in this
instance is thermomechanical resulting in detachment of the coating from the substrate as opposed
to an ablative process dominated by thermal vaporisation. The shock waves from this
thermomechanical effect are audible and can be very loud initially. The intensity of acoustic shock
waves can be used to determine when the coating has been removed since their intensity falls as an
increasing number of pulses are applied to the same area. A plume of ablated material is usually
seen when these short high-energy pulses interact with a surface coating.
The absorption coefficient of a coating for the wavelength of the incident radiation and those of the
substrate are superimposed on the aforementioned processes and if the coating has fillers and other
constituents, these will absorb differently. The absorption coefficient will also be affected by the
colour of the sample and its surface roughness and texture. The coating thickness can also have
effect and this rapidly leads to a very complex multi-variable interaction between the laser and the
coating. Both the reflectivity and thermal diffusivity of the substrate are functions of temperature
and so, taken in combination, a full description of the mechanisms involved in laser cleaning is a
daunting exercise.
In practice, laser cleaning applications have to be optimised for the removal of a specific coating. If
applied to other coatings (with different optical and material properties), cleaning may still be
possible but some operating parameters such as fluence might need to be increased or a lower
cleaning rate accepted. Cleaning rate is largely a function of the average power of the laser output
that has to be high. Lasers used for metal cutting and welding have high average powers because
they are effectively continuous wave lasers. In contrast the high average power of a cleaning laser
requires that the short duration pulse possesses a very high peak power.
For effective laser cleaning it is necessary to remove vaporised material as this might otherwise
absorb sufficient radiation to become ionised and form opaque plasma, effectively shielding the
target from the remaining laser pulse. Usually a filtered vacuum line is coupled to the laser head to
effect efficient removal.
The importance of absorption coefficient (α) has been mentioned. Considering the simple case of
an unfilled resin assumed to possess a single absorption coefficient, if the resin strongly absorbs a
given incident laser wavelength the coefficient will be high. The penetration depth (L), a more
practical measure of attenuation, will consequently be very small (L=α-1). Conversely for a weak
absorber the penetration depth is significantly greater. Roberts14 has studied the interaction of
Nd:YAG and TEA CO2 lasers with epoxy resin and found that L ~ 20 µm for CO2 lasers operating
at 10.6 µm wavelength whereas for Nd:YAG lasers operating at 1.064 µm, L~100 µm.
If the effective penetration depth into the material (at the laser wavelength) is much smaller than the
contaminant coating thickness (d), i.e. L<< d, then removal of the coating by laser interaction is an
ablative process and in the case of a TEA CO2 laser this will involve mainly thermal vaporisation.
However when L~ d, physical detachment of the resin from the shock wave caused by the laser
interaction, will assume a greater role.
In practical terms this means than a TEA CO2 laser will strip a thick layer of contamination more
effectively than a very thin layer and observation supports this. For example an epoxy resin sheet
approximately 150 µm thick was easily ablated after a few pulses from a TEA CO2 laser in the
laboratory whereas a very thin layer cast from solution of the same resin was much more difficult to
remove. In the latter cast most of the energy of each laser pulse was reflected back from the metal
substrate and so many more pulses were required to remove a very thin contaminant layer.
With Nd:YAG lasers, detachment appears to be the dominant mechanism for removal of resin from
the metal substrate. It was observed that a continuous homogenous film of cured epoxy resin (150
µm thick) simply flaked away from the substrate it had adhered to after laser treatment. In other
words the adhesive bonds at the metal/oxide interface were completely broken by the intense shock
wave and the cohesiveness of the epoxy film meant that large surface areas could be removed with
relatively gentle suction. For a less homogenous coating of epoxy resin with granular texture, the
same detachment or spallation process occurs but the removal is less dramatic and stronger suction
is required to prevent ejected material from depositing back onto and adhering to the cleaned
surface.
The absorption characteristics of a metal oxide are different from the parent metal and if the resin is
bonded to a thick oxide layer, removal of the resin will expose the oxide to the energy of the laser
pulse. Using X-ray photoelectron spectroscopy to examine the underside of a flake of detached
resin (resulting from Nd:YAG laser cleaning of epoxy resin bonded to a mild steel substrate) traces
of the metal oxide were detected implying that the oxide is partially ablated once exposed to the
beam energy. The behaviour of the same metal oxide upon exposure to a high-energy pulse from a
TEA CO2 laser is markedly different. In this case the oxide is not shielded from the high
temperature plasma, sitting above the irradiated sample, by the resin, since this has been removed
by thermal vaporisation. Examination of the oxide using scanning electron microscopy shows
innumerable very tiny pits in the oxide surface and it is conjectured that radiated heat from the
plasma melts the metal oxide and that absorbed gases present in the oxide migrate to its surface.
These become effectively frozen into the surface once laser treatment ceases and that rapid cooling
of the molten surface causes micro-cracks to form.
5. Conclusions
A review of the literature reveals that the removal of fully crosslinked material from metal surfaces
cannot be easily achieved using established cleaning technologies. However, both laser ablation
and sodium hydride-based processes can be optimised to achieve this result as demonstrated by the
effective removal of RTM6 from nickel.
Furthermore, an understanding has been gained of the fundamental interaction between different
laser types and epoxide resin and the role of the sodium hydride process in enabling coating
removal.
6. References
1. Sheasby, P.G. and Pinner, R., The Surface Treatment and Finishing of Aluminium and Its
Alloys 6th Ed. Vols 1 & 2, Finishing Publications Ltd, 2001.
2. Cooper, M., Laser Cleaning in Conservation, Butterworth Heinemann 1998.
3. Critchlow, G.W., Cottam, C.A., Brewis, D.M and Emmony, D.C., Int.J.Adhesion &
Adhesives, 1997, 17, 143.
4. Gendler, Z., Rosen, A., Bamberger, M., Rotel, M, Zahavi, J, Buchman, A. and Dodiuk, H,
J.Materials Sci., 1994, 29, 1521.
5. Broad, R., French, J. and Sauer, J., Int.J.Adhesion & Adhesives, 1999, 19, 193.
6. Sancaktar, E., Babu, S.V., Zhang, E., D’Couto, G.C. and Lipshitz, H., J.Adhesion, 1995, 50,
103.
7. Physics World Digest May 2002.
8. Butcher, E., Mould Cleaning with Laser Radiation, Kunstoffe Plaste Europe, 1999, 89.
9. Industrial 2KW TEA CO2 laser for paint stripping of aircraft. G.Schweizer, L.Werner. SPIE
Vol 2502/57.
10. Elbing, F., Anagreh. N., Dorn, L. and Uhlmann, E., Int.J.Adhesion & Adhesives, 2003,
23(1), 69.
11. D.M.Brewis, G.W.Critchlow and C.A.Curtis, ., Int.J.Adhesion & Adhesives, 1999, 19, 253
12. DuPont Patent Specification Oct 16th 1942 no 14502/42 (565,567).
13. Lightfoot, I.P., Reductive Destruction of Chlorinated Organics in Molten Salts, PhD Thesis,
De Montfort University, 2000.
14. Roberts, E., National Laser Centre of South Africa, Private communication.
Figure Captions
Figure1 a. to show areas of RTM6 coating removal on a nickel piece (5 x5 cm) following
TEA CO2 laser ablation. The dark (brown) areas are RTM6 deliberately left intact
for comparative purposes.
b. as a. at higher magnification.
c. Backscatter image of a nickel piece following TEA CO2 laser ablation. The dark
(black) areas are residual islands of RTM6.
d. SEM image of a nickel piece following TEA CO2 laser ablation.
Figure 2 a. to show areas of RTM6 coating removal on a nickel piece (5 x5 cm) following
Nd:YAG laser ablation. The areas irradiated/of detachment are clearly visible.
b. as a. at higher magnification.
c. SEM image of a nickel piece following Nd:YAG laser ablation.
d. as c. with extended treatment and showing evidence of micro-melting.
e. to show the effectiveness of Nd:YAG laser ablation at the removal of a non-
continuous film (lower LH corner irradiated, upper RH corner with contamination
intact).
Figure 3 a. to show an untreated RTM6 coated nickel piece(lower LH corner) and a sodium
hydroxide treated (upper RH corner) treated piece prior to hard rinsing.
b. and c. SEM images of a nickel piece following sodium hydroxide treatment.
d. EDXA scan of a nickel piece following sodium hydroxide treatment.